On 07/11/12 at 07:58pm, Changaco wrote: > Since we got carried away, I'm starting a new sub-thread from the top.
Thank you > Please correct me if I'm wrong, but basically what danimoth is trying > to do is to decentralize what is currently handled by private trackers > in the Bittorrent ecosystem, they enforce minimum share ratios. > > Private trackers aren't perfect for at least four reasons: > - It's harder to find and download the files. > - If the tracker falls all the torrents die. > - Some users try to get a ratio as high as possible, this hurts the > ratios of the other users. > - It doesn't solve the inequalities due to the popularity of the > content. Popular torrents still have better speeds and better chances > of survival. I don't like talking about Bittorrent, because I don't want to assume problems which are relative to an "higher level" (like the popularity of torrents). Here, we could take the discussion on a theoretical point of view, talking generally about "exchanging p2p services for money", which IMHO is really hard, or try to apply these on a real field, which I identified in Kademlia, because of its general purpose scope... it's only a matter of key / values, after all, and then try to abstract the solutions. > Making the mechanism P2P would only solve the first two. > > Furthermore, it might worsen the third one: > - If the solution is based on a generic currency like Bitcoin, which > one can exchange for goods/services or other currencies, the gain from > having an excessive ratio is increased. > - People who have low ratios because others have high ones won't be > able to create another account to start fresh. Ratio is not a problem, IMHO. What I want, as a network architect, is that the network works, and trying to get highest ratio in an honest way isn't bad per se (an example is the GPU fight between bitcoin miners.. they all want the highest ratio as possible, and the network works well under the no 51% assumption). > Maybe a new approach is needed, starting by the exact definition of the > problems that need solving. Free-riders were mentioned, but as far as I > know nobody has proven that they are actually the cause of problems and > what those problems are. > One simple defition that come up to my mind is: design a p2p network which doesn't attract only the effective users of the network itself, but others "casual" users, that share their resources not only to take services from the network itself, but money for example. I could be a miner without using bitcoin which I receive as incentive, but exchanging them with dollars, so in fact I contribute to the network without using its services. _______________________________________________ p2p-hackers mailing list [email protected] http://lists.zooko.com/mailman/listinfo/p2p-hackers
